CancerLinQ: ASCO Dives Into Big Data

Richard L. Schilsky, MD

Today we are discussing the American Society of Clinical Oncology’s (ASCO) health information and patient record initiative called CancerLinQ. A prototype of this system was launched a few years ago to attempt to pool and analyze both real-world and clinical trial data within a single accessible system for oncologists. In November 2013, ASCO announced that it had begun to develop the full CancerLinQ system. We are speaking with Richard L. Schilsky, MD, chief medical officer of ASCO and the former chief of hematology/oncology at the University of Chicago Comprehensive Cancer Center who specialized in gastrointestinal cancers, about this effort, the goals for CancerLinQ, and the progress so far.

 -Interviewed by Anna Azvolinsky 

Cancer Network:Dr. Schilsky, can you tell us about the setup of the initial pilot initiative of CancerLinQ?

Dr. Schilsky: We wanted to build a pilot or prototype for CancerLinQ to prove that it was possible to set up a system like this. We reached out to 30 medical oncology practices scattered around the United States and asked them if they would be willing to provide us with the de-identified records of their breast cancer patients accumulated over the last several years. Our goal was to accumulate records on about 30,000 breast cancer patients. In fact, we collected the records of 170,000 patients with breast cancer, all de-identified. The records came from all 30 of the practices. Each practice contributed a significant number of records. These records were based from four different EHR platforms, which was very important for us to learn whether we could successfully bring in the records agnostic to an EHR platform. From our perspective, the pilot really achieved all of its goals and gave us the confidence that we could move forward with the full build of CancerLinQ.

Cancer Network:What is the impetus for developing the bigger system? What are the issues with the healthcare system that CancerLinQ aims to address?

Dr. Schilsky: The main impetus for developing CancerLinQ is really the increasing complexity of cancer biology and cancer care, and the need for improving the quality of care and assuring the coordination of care. We have a great opportunity to use the electronic record system that is now in place in the great majority of oncologists’ offices around the country to actually learn from the experience of managing each patient with cancer. We have to remember that cancer is increasingly complex at the biological level. Every day it seems we are identifying new molecular subtypes of cancer, each one of which has a different prognosis and requires a different treatment plan. The amount of information that an oncologist has to process in developing a care plan for a patient is greater and more complex than ever before. The needs of patients continue to be very complex and are not necessarily getting simpler, particularly now as we are introducing molecular genetic profiling of patients’ tumors to help guide the treatment planning process.

And of course cancer care, since almost the inception of oncology, has been inherently multidisciplinary, involving medical, surgical, and radiation oncologists, pathologists, and radiologists. Now we are bringing new colleagues into the mix-interventional radiologists who are important for obtaining biopsy specimens that are necessary for the molecular profiling studies, and molecular pathologists to help us do the molecular profiling and interpret the results. Assembling these large teams of specialists who are necessary to optimize the management of individual patients and to coordinate care across various specialty groups is really very complicated.

So, the real impetus for CancerLinQ is, of course, to improve the quality of care that every patient receives and to provide every patient with the opportunity to receive the same level of care, regardless of whether they are being treated in a small community in a rural town or treated at a major cancer center or academic medical center. And we intend to do that by essentially tapping into the information that is locked up in all of the medical records of all of the patients in the country. And through aggregating and analyzing that information, we can develop insights that we can then use to accelerate the process of learning about cancer and disseminate information more quickly and more easily to all oncologists, so that everyone has the most up-to-date information at their fingertips.

Cancer Network:So, is the idea for currently treated patients, for doctors who are part of that patient’s care, to be able to share information more readily, and then additionally, for oncologists to learn and analyze prior patient information to learn how to better treat current patients?

Dr. Schilsky: Absolutely. We will have embedded in CancerLinQ a variety of tools to assist physicians. We are planning to offer, through CancerLinQ, point-of-care guidance and guidelines to physicians so that if they are seeing a patient who has a rare tumor or a genetic abnormality that they are not familiar with, they can essentially call up that information at the point of care through CancerLinQ and get a brief guidance document to help them understand what the best course of action to take will be.

We also imagine that CancerLinQ essentially provides immediate second opinions to oncologists-they will be able to query the CancerLinQ database right from their office desktop by entering the characteristics of the patient they are seeing and get a report back, in real time, about other similar patients in the CancerLinQ ecosystem who have been treated by other doctors-how those patients were treated, what their outcomes were. As cancer becomes increasingly a group of rare diseases, having the opportunities to learn from the experiences of other doctors and other patients will be enormously valuable to each patient and each doctor at every encounter.

Cancer Network:Can you talk about a few of the logistics of setting up the full system that was started at the end of last year, as well as the timelines and maybe a few of the lessons learned from the initial pilot program?

Dr. Schilsky: Let me start with the lessons learned. What we learned from the pilot was several things. First, oncologists are enthusiastic about the opportunity to participate in a program like CancerLinQ, and they are willing to share their electronic health records with a trusted provider like ASCO. Also, that it’s technically possible for them to do that with relatively little burden on the oncology practice, and we can develop a system that can bring in the EHR from essentially any EHR vendor regardless of whatever platform the doctor is using in their office. We learned that we are able to map that EHR data into a standard format and aggregate it, de-identify it, and ultimately learn from it.

In the prototype we were able to pose a number of clinical questions to the CancerLinQ dataset that had been assembled and see what kind of answers we got. And in most cases, the answers we got were completely consistent with what we knew to be the truth from the results of completed randomized clinical trials or other pieces of clinical information that verified the observations we made in CancerLinQ. The lessons learned really provided the proof of concept that we needed: that oncologists would participate, that it was technically feasible to build, and that it could provide valuable information.

That then led the ASCO board of directors, almost a year ago, to give their approval to the full build of CancerLinQ and to commit substantial resources to achieving that goal. We then spent a good deal of time last fall gathering the functional requirements for CancerLinQ. That involved a lot of consultation with potential user groups so we could understand from the perspective of doctors, other healthcare team members, and patients-how they would find value and what they would use a tool like CancerLinQ for, what they would want it to do, how they would want it to work. And that enabled us to put together a long list of the functional requirements that would need to be met to build CancerLinQ successfully. That then led, in April of this year, to ASCO issuing a request for proposals to the health IT vendor community to assist us in the technical building of CancerLinQ. We got a large number of responses, and we are still in the process of deciding the final plan for exactly how CancerLinQ will be built and what entities we will be working with to move forward with the final system.

We have a commitment to the ASCO board of directors to present them with a recommendation by August, in a few weeks. We expect that they will approve the plan and then soon thereafter begin the formal technical build of the system. The goal is to have the system deployed, on a limited scale, by the third quarter of 2015. And we already have 12 practices around the country that we call our vanguard practices, who have indicated their willingness to be the beta sites. They will be the first practices to receive CancerLinQ, first to submit data, and the first to receive the many reports that CancerLinQ will provide. By a year from now, we will have a lot more to talk about!

Cancer Network:Are there any types of regulatory approvals that are required for the transfer and access to patients’ clinical records that are pooled?

Dr. Schilsky: Let me briefly explain the regulatory framework that we recently published in a paper in the Journal of Clinical Oncology. There are two major sets of laws that we need to pay attention to-one is HIPAA and the other is the Common Rule. Under HIPAA, doctors are permitted to engage with a business associate, under what is called a business associate agreement, for that business associate to perform certain quality and business analytics on their medical records for the purpose of improving the quality of care. And that is a permissible activity under HIPAA and does not require the formal consent of the patients, for their medical records to be used in that way. CancerLinQ will be implemented under HIPAA compliance of business associate agreements and that will give the practices the ability to transfer the electronic medical records to us for the purposes of quality analytics.

The Common Rule governs human subjects research-technically, it governs federally funded human subjects research. So, on that technicality, CancerLinQ does not fall under the Common Rule because it is not currently federally funded. But the important thing is that human subjects research that is done using de-identified information is not subject to the Common Rule and does not require a patient’s informed consent.

The ultimate analysis of CancerLinQ is to obtain the records for quality improvement without formal patient consent, to use the information in CancerLinQ once it is de-identified, and to address various research questions that can also be done without patient informed consent. That said, we believe that we have an obligation to inform patients about how their medical information is being used and for what purposes, so we will develop a disclosure process so that patients are aware and informed that their doctor is a participant in the CancerLinQ program and that their records will be contributed to the system for quality improvement and a variety of other purposes.

Cancer Network:Thank you so much for joining us today, Dr. Schilsky.

Dr. Schilsky: Thank you.